A collaboration between EBI and Google Research has annotated ~49 million previously uncharacterized proteins that are
now part of the UniProt database.  The annotations are based on a Protein Natural Language Model called 'ProtNLM'.

Three datasets were created to facilitate use by the maize research community:

---------------------------------------------
Uniprot_ProtNLM_maize_annotations.txt

A tab-separated file filtered for maize genes.  The original annotation file
(https://storage.googleapis.com/brain-genomics-public/research/proteins/protnlm/uniprot_2022_04/evidencer_sorted.csv.gz)
was filtered for only annotations of UniProt IDs with B73 RefGen_v5 gene models at MaizeGDB.
---------------------------------------------
Uniprot_ProtNLM_maize_v5.txt

A tab-separated file for B73 genes and annotations with the following columns:

B73 RefGen_v5 gene model ID
Previous UniProt description
New ProtNLM description

---------------------------------------------
Uniprot_ProtNLM_maize_v5_uncharacterized.txt

A tab-separated file for B73 genes with previous "Uncharacterized protein" annotations with the following columns:

B73 RefGen_v5 gene model ID
Previous UniProt description
New ProtNLM description
---------------------------------------------

ProtNLM description:
Website: https://www.uniprot.org/help/ProtNLM

Gane, A. et al. ProtNLM: Model-based Natural Language Protein
Annotation (2022) Preprint: https://storage.googleapis.com/brain-genomics-public/research/proteins/protnlm/uniprot_2022_04/protnlm_preprint_draft.pdf

Last update 10/13/2022 - C. Andorf